In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss how to leverage large language models like ChatGPT with your business analytics data. We talk through real examples of using AI to analyze CRM data for buyer personas, extract tables from PDFs, summarize PowerPoints into executive briefs, transform your data into actionable insights, classify social media comments, and answer questions about your data. We explain that while generative AI is exciting, your analytics data should still be the backbone of business decisions. Tune in to learn creative ways analytics and AI can work together to enhance your marketing strategy.

Subscribe To This Show!

If you’re not already subscribed to In-Ear Insights, get set up now!

Advertisement: Generative AI for Marketers

Imagine a world where your marketing strategies are supercharged by the most cutting-edge technology available – Generative AI. Generative AI has the potential to save you incredible amounts of time and money, and you have the opportunity to be at the forefront. Get up to speed on using generative AI in your business in a thoughtful way with our new offering, Generative AI for Marketers, which comes in two flavors, workshops and a course. 

Workshops: Offer the Generative AI for Marketers half and full day workshops at your company. These hands-on sessions are packed with exercises, resources and practical tips that you can implement immediately.

CONTACT US

Course: We’ve turned our most popular full-day workshop into a self-paced course. The Generative AI for Marketers online course is available now!

REGISTER NOW

Key takeaways you’ll receive: 

• Hands-on learning for the six major use cases. When you’re done with the course, you’ll have your first collection of working prompts based on your business and needs 
• A plan and sales pitch to help you sell the use of AI internally to stakeholders 
• A roadmap for who should be on your AI pilot team and your internal AI council 
• Your first custom AI that you will build for your business (paid version of ChatGPT required) 
• Tactical and strategic knowledge of generative AI to help you lead your team in the new AI landscape 
• A priority matrix of the tasks you’ll want to consider for AI enablement

Sponsor This Show!

Are you struggling to reach the right audiences? Trust Insights offers sponsorships in our newsletters, podcasts, and media properties to help your brand be seen and heard by the right people. Our media properties reach almost 100,000 people every week, from the In Ear Insights podcast to the Almost Timely and In the Headlights newsletters. Reach out to us today to learn more.

Watch the video here:

https://youtu.be/pgGNObybh3A

Can’t see anything? Watch it on YouTube here.

Listen to the audio here:

Download the MP3 audio here.

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

 

Christopher Penn 0:01

In this week’s In-Ear Insights, we have been talking and everyone has been talking about large language models, the tools that power things like ChatGPT, for ever and ever.

And one of the things that we want to be cognizant of, is not losing sight of the fact that the workhorse of machine learning and AI is typically around good old, almost boring analytics data.

How do we use these these newfangled fashionable tools with data that still runs our businesses? So Katie, when you think about analytics, and what’s the what are the kinds of data that you might want to be able to ask a language model about? And then we can talk through some of the use cases, I did a talk recently with our friends over at martec martec.org, about using language models with analytics data?

Katie Robbert 0:50

Well, I’m glad we’re talking about analytics data, because I feel like large language models and generative AI have overtaken the conversations in a way that we have been distracted and almost forgotten.

That things like our analytics data are what power the business are what help us make decisions, and that without it, these other pieces, the shiny objects are just that.

So when I think about analytics data, you know, obviously, I think about my website data, things like a Google Analytics, or Adobe analytics or a matomo, telling me what’s happening within my website in that ecosystem.

But then I also think about like my CRM data, what are people opening and what are people ignoring and what’s causing someone to unsubscribe or decide that they want to stick around.

And then the third piece is my sales and revenue data.

So those would be the three pieces that I would look at.

I guess, when I talk about website data, it’s, there’s a bigger bucket, there’s the actual digital marketing data that a lot of different things go into sort of the marketing automation, our email data, there’s our website data, there’s social data, there’s a lot of other pieces.

So all of the things that tell me what’s happening with our digital footprint, is what I care about.

And then also my revenue.

Christopher Penn 2:18

That makes good sense.

In fact, in the the talk we gave last week about this stuff, we talked specifically about some of those things, so much so that we actually did some demonstrations with synthetic CRM data, and what so there’s sort of six broad categories that we talked about all the time with language models, generation, aka making stuff, extraction, getting data out of things, summarizing stuff, rewriting stuff, classify, classifying stuff, and then answering questions about stuff.

And CRM data is one of those pools of data that is incredibly valuable, very rich, and no one uses it, no one uses it.

And they should,

Katie Robbert 2:59

they should.

You know, I remember Chris, probably about this time last year, you and I went through an extensive requirements gathering phase of how to build out more sophisticated reports from our CRM data.

This was prior to systems like Claude and ChatGPT, hitting the mainstream market.

So you know, maybe in a different episode, we can talk about how our approach to those requirements or approach to the execution might have changed.

But that’s exactly it.

We are not using our CRM data to its fullest extent because of the reporting limitations from the system itself.

But I’m wondering now, like, what could we do using some of these large language models.

Christopher Penn 3:44

And that’s, that’s exactly kind of what we showed.

And I’ll just show a very quick demo here just to give people a sense of of what you could do.

So in ChatGPT, the paid version under GPT-4, this tool called Code interpreter, which is horrendously named, it really should be called Junior data scientist, I’ve taken our CRM, or I’ve taken our synthetic CRM data I want to emphasize this is completely manufactured data is not real data.

And what I’ve asked it is, I want you to help me build a buyer persona based on all the characteristics of this dataset.

And tell me what the closed one deals have in common that they don’t share with closed last deals, so that I can try and figure out like, what is who is my buyer? And I think, you know, particularly for companies have a lot of data.

This is a very valuable exercise to go through.

Katie Robbert 4:44

We’re always chasing like who that ideal customers who are they what do they want, what are they thinking? And the challenge was that with the without using a large language model is that we You don’t necessarily gather enough information about someone in order to create that profile.

And so I would imagine with the support and assistance of a large language model and some machine learning, you know, data inference, we could actually start to build out, you know, what we typically call personas.

And those profiles, and then do different kinds of segmentation, and get recommendations on what to do with that information.

Christopher Penn 5:29

Exactly.

So what in this case the The GPT-4 tool is doing is it is writing its own code to do the kind of the kind of analytics to say, like, I need to understand the data set First, I need to to build the appropriate code and then do things like the state will suggest the statistical tests to run, should you run a z score? Should you run a chi squared, test? Should you run a logistic regression and so on and so forth.

And, again, I’m not going to spend a whole time of time going through this here.

But that’s one of the six use cases for analytics data with language models is because tools like ChatGPT, and Claude to to some extent can write their own code and test it and make sure it works.

You can have it do very sophisticated analysis of any tabular data.

So CRM data, marketing, automation data, Google Analytics data, you name it, if as long as it fits within the the allowed file size, you can have it do those experiments, and then say, Okay, well, who is our buyer? Like? What are their patterns in the data that we just can’t see, but the software could?

Katie Robbert 6:33

Or what’s my, the, where my brain is going with all of this is? It’s one more system to manage.

And so I know, there’s a lot of connectors, and a lot of companies that are building an interface on top of systems like ChatGPT.

So, you know, if you brought this to me, Chris, and said, I can analyze our CRM data, my first question for you is like, okay, but where does that live? Do I have to keep going back and forth between the CRM and the large language model? Or do we have a more centralized place to do this where I can, you know, hit the magic buttons and say, you know, reanalyze.

Christopher Penn 7:19

And this is where the governance portion of these tools, you know, martec governance, because that’s really what we’re talking about, is so essential, because, as you saw, in the very, very quick demo, it’s writing Python code, it is literally just writing Python code, and even a moderately skilled Python person could copy and paste the steps that it takes, once it’s done, assuming you’re satisfied with the exploration, and you would then put that in production somewhere within your enterprise, and say, Okay, now here’s, here’s the data set, now push the button, and it’ll rerun that entire analysis the next time through, so you don’t have to have it, you’re gonna have a person doing the interrogation, you can now have come up part of your process and part of your internal platform.

And so that’s where that’s where this goes from cool thought experiment to, okay, this is a usable processing system within our company.

Katie Robbert 8:14

And I think that that sort of like, when I think about gender, generative AI still being a lot of a shiny object, those are the steps that aren’t yet being taken by a lot of companies.

Everything is still living in existing within the generative AI system.

And so these use cases for analytics data, they are they’re sort of that novelty of like, it’s cool, I could do this thing.

But people like me or SMS would like, but how do I streamline it? How do I make it? So it’s not yet one more piece of equipment in my tech stack?

Christopher Penn 8:49

And to some degree, it is, I would, I would say you think of it like your r&d lab almost.

Right.

It’s it is sort of your, your r&d lab where you, you test stuff out, and then once you find things that work, then you figure out, Okay, what’s the process internally to move that those things into production?

Katie Robbert 9:10

That makes sense.

Alright, so we talked about the CRM data, which I think is super useful.

And you and I are going to catch up about that offline because I have a lot of questions and ideas.

What’s another use case that we are overlooking with our analytics data?

Christopher Penn 9:26

extraction of data from other data sources is the second one.

So this is clods and Anthro pics Claude, I’m uploading a PDF that contains a lot of tait tabular data.

Now, anyone who’s ever worked with PDFs knows getting data out of PDF Oh is a delight.

And by the way, you academia and in fact, this is an academic paper.

So I’ve given it I’ve given it a prop say, I want you to try and extract the tables.

Now.

This prop sometimes works sometimes does not depends on on the file and how Plot A feeling that day.

But in this case, it is, in fact, actually extracting out several of the tables in a pipe delimited format.

So if you have data that is locked away in a PDF, you can get extracts from it, which is, again, very helpful, very straightforward use case.

But if you’re sitting around with lots and lots of data, or one of my favorite use cases is, maybe you have that person in your company who loves to make ridiculously complex Excel spreadsheets and you just want a set of rows, I ran into this with some of the Warrior Nun stuff, one of my friends in that enum has this very bizarre format for their data, like I just want a table and this is broken up into all, you can give it to a tool like Claude or ChatGPT.

And say, please just extract this in a rectangular table.

Katie Robbert 10:52

I can also think of a lot of, to talk about offline, a lot of uses for that, as well.

This, as an aside, one of the things that makes me nuts are people who use Excel spreadsheets like Word docs, and then all there’s all this other stuff in the sheet, not just metrics, and then I can’t sort it, I can’t do it, I can’t filter it.

Because there’s all these words and pictures.

I’m like, this is a sheet.

And so being able to translate that using one of these large language models is, for me, invaluable.

Short of just slapping someone’s hand and saying, don’t touch that.

Christopher Penn 11:33

Exactly.

The third use case is summarization.

And this is something that I was absolutely lost without.

And now it is something that is part of every every month I do this in our reporting, and it saves me so much time, the PowerPoint, as when you have like little call out boxes, and notations and stuff in PowerPoint.

And there’s, those are actually data fields.

And for the longest time, like how do I get the text out of a PowerPoint document, so I can make a good executive summary, and not gonna say anything and make some action items.

And so I started doing a little bit of research then like, I’m just gonna ask ChatGPT Hey, what format is this in PowerPoint, and it turns out that, because it’s a big model, it knows what that format looks like.

And so I wrote, it wrote me some PowerPoint, some Python code to say, find all the text in this presentation, and export it as a single text file.

So I ran this recently on our on the Trust Insights, SEO report.

And of course, it spits out all the the notes that I made throughout, you know, all little call out boxes throughout the presentation.

And then I take that, and put it in a language model and say, summarize this into executive summary.

And that’s what you see every month when I when I share it internally.

By the way, if you are interested in that Python code, and you would like a copy of it, it is available.

If you go to our analysts for marketers slack group, I shared it recently I said, Hey, here it is, anyone who wants it for free.

But you got to be a member of analytics marketers, you can go get it trust insights.ai/analytics.

For marketers.

Katie Robbert 13:09

How do you know it’s writing Python code? Does it say this is Python code?

Christopher Penn 13:15

It does.

And I also recognize the syntax.

What and so

Katie Robbert 13:18

that’s that, you know, for someone like me, I would look at that and go, I don’t know what this is.

And so there’s definitely a learning curve for using these large language models.

And some of these use cases.

So you know, I would, I would know who to ask, or I would know how to research.

What is it that I’m looking at.

But for some people, if a large language model just starts writing code, they’re gonna look at it and go, I don’t know what to do with this, or is it just throwing, you know, junk and errors at me? You know, so it’s interesting that this is the response that it’s giving you like, so could I give it my data, my analytics or my CRM and say, Help me put together a solution that doesn’t require writing code?

Christopher Penn 14:06

Maybe that’d be, that’d be tricky, because because it’s not accessible for everybody.

It’s not accessible for everybody.

But these tools are not mathematical tools.

And so one of the misconceptions about them is they are Language Tools, hence the name lodge language model.

And so they sort of sidestep their lack of mathematical skill by writing code, because code is still a language, and then it executes the code.

So I would say in this particular use case, there, I mean, the easiest way to do that would be for us to turn it into a web app.

And then, you know, offer it to clients and say, like, Hey, if you want to use this upload, your PowerPoint presentation will spit back the thing and it’ll be a web based service instead of something that people have to run locally if they don’t have those technical capabilities in hand.

else.

And if you’d like to let us know if this is a service that you would be willing to pay for hop in our Slack group or let us know, by going to trust insights.ai/contact.

Katie Robbert 15:12

So that’s an interesting, you know, sort of pro tip is that these large language model, the solution might be, it’s going to give you back pieces of code to then run.

And so I think that’s one of the misunderstand, it’s your point is one of the misunderstandings about how these large language models are putting together solutions for you.

So, so we’ve talked about CRM data, we’ve talked about extraction summarization, what were some of the other use cases that we are overlooking.

Christopher Penn 15:48

The third one is what I call transformation of data.

So transformation of data is when you take an existing data that you have, and you asked for it to be converted into a different format, a format where you can, you can ask it to interpret things, you can ask it to, to compare things.

So here is an example.

I’m going to paste in this relatively long prompt I have about Google Analytics date, and I have some Google Analytics session data in here.

And I’m asking it to help take a look at it.

So it’s going to start doing the analysis, it’s going to be run, essentially, come up with the computations and needed to compare these two periods and Google Analytics, I do appreciate the fact that it actually puts the mathematical formulas there.

And it’s going to spit out essentially, its findings from our Gulags data, and then what recommendations it should make.

So this is an example of a rewrite where we’ve given an existing data and we’re asking you to rewrite that data in a different format in the, in this case as, as recommendations.

Katie Robbert 17:00

Interesting.

And so it’s going to go through this whole, you know, code rewrite.

And then for those listening ChatGPT, right now has taken what Chris has given and is rewriting the code, and it’s now oh, it looks like what it’s now also giving us is the actual analysis of the code that it’s written.

And so it’s an interesting way to look at because you didn’t give ChatGPT code to start with, you said, this is the thing I want to do.

Here’s the data.

So the solution that this large language model came up with was to write its own code in order to understand how to analyze the data in the way that you’re asking.

Christopher Penn 17:42

Exactly.

And so in the observations, it says here, the direct traffic channel saw a significant decrease by 34%.

This is a good sign as this can mean, the tagging has improved.

And there’s better data quality, organic search traffic increased slightly, which is just the SEO efforts might be paying off organic social saw a drastic decrease of 46%.

This may suggest the social media campaigns, your content wasn’t as effective in the current period where there might be traffic coming from dark social, which is being tracked.

Now, it’s also going to start doing some recommend campaign recommendations.

So like if your email segment the email list and target users based on on preferences, direct traffic, keep doing good tagging, and so on, and so forth.

Now, these are, these are pretty vanilla recommendations.

But it’s a good example of, I’ve got some data.

And maybe I need to write a report for my stakeholders, and I’m just stuck.

Or maybe I didn’t have, you know, that second cup of coffee that morning.

This is a good starting point, a good first draft of what you could be doing in your analytics reporting.

Katie Robbert 18:42

And I think that’s, that’s the key is tools like ChatGPT, and Claude are good, jumping off places, they’re good to help you get unstuck.

But you shouldn’t take this verbatim, paste it into a report and give it to your stakeholders.

Because your point President BERT in ELA, because your business, your campaigns, your digital marketing is nuanced.

And there’s information in here, it’s just looking at session data.

So it doesn’t know all of the other outside factors as to why your session data may be so you know, for example, it said that organic social wisdom, I can tell you exactly why that is.

It doesn’t know the reason but I know the reason why organic search lives down.

You know, if if this were our data, I would say because we have pulled back resources for posting on social that we still use organic, social, but not in the way that you were six months ago, 12 months ago.

And so that’s information that this system doesn’t know.

Christopher Penn 19:44

Yeah, this is actually not our data.

This is from the Warrior Nun campaign.

But yes, exactly.

But

Katie Robbert 19:48

yeah, it’s as an example, like I could say, I know why referral traffic is down but this system wouldn’t necessarily know

Christopher Penn 19:55

exactly.

And that’s context that you can and should provide in your prompt them That’s an important part.

The next example is classification.

So this is taking data and classifying.

So I have here is a synth, a bunch of synthetic reviews from a Norwegian restaurant, hang on here to get file loaded.

do reviews.

Katie Robbert 20:18

And so where do you load the data in these systems? Is it through?

Christopher Penn 20:26

Oh, right at the bottom there, this is Claude.

Okay, this is close little Attachments button.

Katie Robbert 20:31

But there isn’t an attachment button in ChatGPT.

Christopher Penn 20:35

There is if you use code interpreter.

So regularly ChatGPT does not have that the code interpreter does.

But code attracts the paid version.

Yes, code interpreter can only take tabular data, CSV files, text files, et cetera.

So it can’t take in like images, or movies and things like that.

So here we have clause assessment of 5000 records.

This these are customer reviews for the Arendelle cafe, a mixed rating of 2.9 strengths, authentic Norwegian cuisine, weaknesses, inconsistent food service recommendations, improved training for waitstaff, customer reviews, customer feedback, Voice of the Customer stuff, this is analytics data that we typically have in abundance, but because typically unstructured, we don’t use it.

I mean, Katie, you and I have talked about this with many clients saying, like you’re collecting the data, you’ve got an email inbox, it’s overflowing with customer feedback, and you’re not using it, why not? And right answers.

We don’t know how.

Now there’s a way

Katie Robbert 21:37

I’m thinking of a couple of our clients who we should probably proactively do this for, because those are the questions like How, why are people you know, not subscribing or signing up or giving us money? Well, we have a lot of that information as to why

Christopher Penn 21:52

exactly.

You’ll recall a few years ago, we had to build an incredibly complicated network graphing system to help one of the clients identify a potential new products in the marketplace using some of their data.

That project today, that project took three months just to write the code for and to get it working.

I could do the same project today in about 25 minutes with a language that because for that particular data, that data is language itself customers words, a language model is the right solution for that much better than what we came up with back then.

Because this didn’t, this wasn’t available to us back then.

Katie Robbert 22:33

Give me a lot of ideas today.

Christopher Penn 22:36

Oh, good.

That means more work.

And the last category is question answering.

And again, this is where once you’ve got the data set in here, you can start asking questions of it.

You can say, what significant anomalies do you see in the data? Are there particular standout extreme extremes that are worth analyzing further? And the critical thing to remember about question answering when your question answering with a language model is you don’t need to have technical subject matter expertise, you just have to have your business subject matter expertise.

So if you are the cafe owner, and people are complaining that the gear toast, the Bruno’s is is not fresh.

And you know that it takes two and a half months to get it shipped from Norway to your restaurant, you’re like, Okay, that’s a complaint, but I can’t do anything about that, right? Because that’s so I can’t address that prompt.

So you would have subject matter expertise of your business you have so here it’s going through this synthetic now the reason is coming up with these these anomalies is because this is completely synthetic data.

So it’s reused the code I used to generate it, reuse the same names for a lot of things so that those that’s expected.

But this shows I think this is important.

The O duplicate reviews by Hiroshi lack of responses to some batteries and contradictory to details and others point to some amount of review manipulation or fabrication.

Okay, so now, if you are a B2C company, maybe you’ve got a product on Amazon, you’ve got 5000 reviews, maybe you can start analyzing going Are we being reviewed bombed? Or are we being manipulated by a competitor or something? If you are a consumer, even you could take take the reviews on a product, like how many these reviews are real.

There’s a product, there’s a big ticket product you’re thinking of buying, like maybe this would be a useful technique to understand that

Katie Robbert 24:40

it’s interesting you bring that up because as a consumer, I always look at reviews for bigger ticket items and a lot of them feel planted by the company themselves.

Or you know, I know I’ve seen you know the site muscle make $1,000 You know, reviewing Amazon products for our like, okay, but how’d you get your hands on all As products in order to renew them, like there’s a lot of, you know, buyer beware lessons that, you know, regular people who aren’t running businesses could be using these large language models or just for their own research.

Christopher Penn 25:17

So let’s ask, is there evidence that these reviews might be faked or made by the company.

So let’s look at the text of the reviews because I think that’s, you know, if you’re gonna spend two grand or three grand on a big ticket item, you might want to know that you don’t necessarily need to have a fancy scraper or anything like that, if you if you’ve got enough of a suspicion, you could just copy and paste into a text document page after page if it would take you a little while.

But if it takes you 15 minutes, just copy and paste a bunch of reviews.

But it could save you three grand on a mistake that might seems like it’s something worth doing.

And then of course, you can load that into a system like Claude or ChatGPT.

And just ask it Hey, tell me about the language of these reviews.

Here it’s saying overwhelming praise and any five star reviews without any critique.

Most genuine reviews will mix good and bad points, even overall positive.

Details like menu items, chefs dedication, kids loving pancakes, things like inside knowledge of the restaurant operation rather than average customers experience.

Now of course, again, this is a synthetic data.

Katie Robbert 26:16

But oh, that does make a lot of sense.

Like how would someone who only went there once No, this

Christopher Penn 26:23

occasional awkward phrasing also hints at non native marketing writers rather than genuine customers.

I think there’s some really interesting language in here.

pattern was specific menu item pairs described as amazing combinations like let’s say plus Jampol, smoked salmon.

Most customers wouldn’t necessarily know or care about these pairings unless suggested.

So again, you could decompose a lot of content out there go hmm.

Is there something suspicious about that? So these are the six big use cases for large language models with analytics data.

And I think it’s really important to reiterate your analytics data is the backbone of your operations, it is the it is your decision making tools, these these language models and their interfaces give you obstetrical if you are a non technical non data marketer, a window into asking your your data, the questions you want answers to, but

Katie Robbert 27:24

I like it.

And I think you know, the, you know, thing to reiterate, is having an understanding, having an expertise in your business is important when asking these questions, because you need to know if the responses you’re getting are valid or not, or are worth, you know, digging deeper into.

Because if you just hand it a set of a data set that you know nothing about, yeah, it’s gonna give you information.

But how do you know whether it’s valid or not? So, you know, if I’m looking at data on the plumbing industry, well, I’m not a plumber, I don’t work within that industry.

So I have no idea if you know, the cost of valves going up or down is a good thing.

Christopher Penn 28:10

Exactly.

But you can also use the tools.

And we’ve talked about this in past episodes as professional development tools.

If you need to get up to speed on something on a particular subject, you can absolutely use them for that to have you almost tutor you in a way of exploring.

Okay, this is how this thing works.

This is where this is what this concept is, is this valve in plumbing important.

What’s the supply of it? I saw on the news this morning, there’s a significant shortage of rice on the planet right now doing it for a whole bunch of reasons.

And if you’re an economist that that makes sense.

And you can see the big picture, if you are not an economist, you might not know what that means.

But you can ask a language model, Hey, I saw this analytics data about this world’s supply of rice.

What does this mean for me? How what should I be doing about thinking about it? How would How should this affect my grocery purchases? Why is the other things at the store increase in price? If you’re a business owner? Like how will that change your business? That seemingly, you know, agricultural product will affect Trust Insights in some way, because the price of consumer staples means that there’s less money consumers have to spend elsewhere, which means they spend less money with their vendors.

Those vendors and transpose spend less money with their vendors.

And that eventually works its way through the supply chain to companies like us.

We’re like, okay, it could turn out that that rice shortage could have a spillover effect on two quarters from now when companies like yeah, we got our budgets cut the

Katie Robbert 29:37

butterfly effect.

It

Christopher Penn 29:38

is and the thing about language models, particularly the big ones like GPT-4 is they can see through our language, very, very big picture of the world and draw those connections that we can’t and it’s why it’s so powerful to use these winner analytics data.

Katie Robbert 29:56

I like it.

I like it.

I have a lot of I have a lot of thought.

Some questions and ideas so, you know, get your pen out or get your otter transcriber out for our next conversation.

Christopher Penn 30:09

I certainly will.

And the summary of these use cases, we actually actually have them written down along with the prompts if you would like them, again, go to our free slack group go to TrustInsights.ai AI slash analytics for markers.

We’ll put it in there.

And you and 3300 other marketers can ask to answer each other’s questions all day long about analytics, also about AI.

We’re always happy to have folks having conversations and also things like you know whether pineapple belongs on pizza or not.

The answer is yes, by the way.

It does not.

Katie Robbert 30:39

No, you will never change my mind.

Christopher Penn 30:44

And wherever it is, you watch or listen to the show.

If there’s a channel you would rather have it on instead, you can find it at trust insights.ai/ti podcast where you’re on most channels.

Thanks for tuning in, and we will talk to you next time.

---

Need help with your marketing AI and analytics?	You might also enjoy:
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday! Click here to subscribe now »	Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday.